artificial intelligence, artificial life, artificial agency, artificial physics | @martinbiehl@mathstodon.xyz

tokyo/berlin
Joined July 2012
I feel pretty dumb for never even having considered this. Roughly: Neural network recursive self-improvement speed must be limited by neural network training speed. Unless there is some faster non-neural network tech that the neural networks discover of course. Still interesting.
Replying to @jeffcafe_
yea but algorithmic foom on neural networks wont foom as much as algorithmic foom on garage-brewed, i386-ran Magic Intelligence Formula
2
1
10
1,051
I am happy that many who vote red note it is a positive thing that if everyone voted red everyone survives. It suggests to me that far more than 50 % of people prefer nobody would die as a result of this.
Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?
42%Red
58%Blue
98,539 votes • Final results
110
At what point do you think the experience of watching a sunset was first modified? Feel free to share your reasoning 🙏
44%Step 1
16%Step 2
2%Step 3
38%It never changed
50 votes • Final results
16
1
2
1,494
martin biehl retweeted
Starting Thursday, we'll be updating our revenue sharing incentives to better reward the content we want on X: We will be giving more weight to impressions from your home region—to encourage content that resonates with people in your country, in neighboring countries and people who speak your language. While we appreciate everyone's opinion on American politics, we hope this will disincentivize gaming the attention of US or Japanese accounts and instead, drive diverse conversations on the platform. We invite creators to start building an audience locally. X will be a much richer community when there's relevant posts for people in all parts of the world.
Readers added context they thought people might want to know
Elon Musk has put this change on hold. “We will pause moving forward with this until further consideration” x.com/elonmusk/statu…
10,331
3,535
8,146
36,257
16,425,780
Replying to @littmath
I think there's a real tradeoff here -- for a high percentage of real ChatGPT questions (even many in math), you really want it to ground everything with search rather than thinking for itself.
1
1
12
666
martin biehl retweeted
Found this game! (thanks to @HirokiSayama) It turned out to be github.com/timhutton/livingp… by Tim Hutton and it's quite captivating
🧬🦠🧪Could you help me find a puzzle game I played a few years ago and forgot it's name? It was tiny and web-based, I guess on someone's personal website. Colorful particles were floating and colliding in 2D space and reacting (binding, unbinding or changing colors) according to simple player-defined rules. The objective was to design a set of rules to build some structure, like a chain or a membrane that would only allow some type of particles through it. Does it ring the bell?
2
8
89
7,983
2/ Goal-Directed behavior in living and non-living systems II focuses on what it means for a system to display goal-directed behavior (GDB) , emphasizing how GDB can be described using tools from information theory, thermodynamics, and Bayesian statistics.
1
1
186
#POSTDOC avaiable at @istecnico. Combining modelling chemical networks, goal-directed behavior, & #OoL. Good conditions, but needs to be filled ASAP by a person who is already allowed to legaly live & work in Portugal. Check full details at: euraxess.ec.europa.eu/jobs/3…
4
8
1
16
4,529
Just wanted to check if AI is still dumb and now I need a drink. I asked the leading AI websites to generate an image of a FULL glass of wine and this is how it went... Chatgpt:
171
161
61
5,432
852,730
At #TAIS2024 @36zimmer formalised the notion of agency for Moore machines. He shared his insights into detecting agency within this context and speculated about its broader implications. Watch now: youtu.be/wPo29I1OGcY #AIsafety
2
3
414
I particpated in a great panel discussion at @crosslabstokyo with @36zimmer, @kaixhin and Paul Colognese, on "what is an agent?" "what's the relationship between goals and beliefs?" and "where do we draw the boundary between an agent and its environment?" youtube.com/watch?v=lEzJfJsX…
1
5
1
23
2,839
📜✨ New paper out! In this jww @u_map_prop and David Spivak we introduce 'open energy-driven systems', which are a broad class of differential systems which evolve by generalised gradient flow
Matteo Capucci, Owen Lynch, David I. Spivak: Organizing Physics with Open Energy-Driven Systems arxiv.org/abs/2404.16140 arxiv.org/pdf/2404.16140
3
28
3
133
21,834
New preprint, comments very welcome. I argue from basic principles of thermo that Jeremy England's "Statistical physics of self-replication" doesn't provide a thermodynamic bound on replication. In fact, I show that no such bound is possible in principle arxiv.org/abs/2404.01130
3
30
4
125
20,866
Would you like to learn more about #GoalDirectedBehavior? Or, would you like to brainstorm your ideas with a larger audience?. Then join our workshop this #July, to be held during the 2024 @ALifeConf. Check out: sites.google.com/view/gdb-al…
2
12
40
2,540
I'll be giving a talk on agency at tais2024.cc in Tokyo on 6 April 2024. The conference is free to attend in person or online.
5
18
2,590
We are excited to have started an international project about #GoalDirectedBehavior funded by @templeton_fdn! We aim to develop a theoretical framework for Goal Directed Behavior and apply this to systems ranging from Chemical Networks to motile life-like Oil Droplets.
3
24
1
150
17,908
martin biehl retweeted
Conway’s Game of Life is Omniperiodic arxiv.org/abs/2312.02799 “In the theory of cellular automata, an oscillator is a pattern that repeats itself after a fixed number of generations; that number is called its period. A cellular automaton is called omniperiodic if there exist oscillators of all periods. At the turn of the millennium, only twelve oscillator periods remained to be found in Conway's Game of Life. The search has finally ended, with the discovery of oscillators having the final two periods, 19 and 41, proving that Life is omniperiodic. Besides filling in the missing periods, we give a detailed history of the omniperiodicity problem and the strategies used to solve it, summarising the work of a large number of people in the decades since the creation of Life.”
5
126
8
566
70,724
Algorithmic information theory (AIT) defines information as description length (not statistical uncertainty, as in Shannon's IT). In new work, we combine AIT + stochastic thermodynamics to uncover a new "algorithmic cost" for thermodynamics of computation journals.aps.org/pre/abstrac…
3
17
3
112
11,677
🔥Welcome to our workshop "The Distributed Ghost: Cellular Automata, Discrete Dynamical Systems, and Their Application to Intelligence" at @ALifeConf #ALIFE2023. Date/Time: July 27th, 2023 at 10:10-11:30 and 13:00-14:20 in Sapporo (Japan) Program 👉 nichele.eu/ALIFE-Distributed…
2
1
8
2,878